National Academies Press: OpenBook
« Previous: Chapter 2 - Existing Studies Using In-Vehicle Video Cameras
Page 15
Suggested Citation:"Chapter 3 - Dimensions of Data Feasibility." National Academies of Sciences, Engineering, and Medicine. 2011. Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion. Washington, DC: The National Academies Press. doi: 10.17226/14509.
×
Page 15
Page 16
Suggested Citation:"Chapter 3 - Dimensions of Data Feasibility." National Academies of Sciences, Engineering, and Medicine. 2011. Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion. Washington, DC: The National Academies Press. doi: 10.17226/14509.
×
Page 16
Page 17
Suggested Citation:"Chapter 3 - Dimensions of Data Feasibility." National Academies of Sciences, Engineering, and Medicine. 2011. Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion. Washington, DC: The National Academies Press. doi: 10.17226/14509.
×
Page 17
Page 18
Suggested Citation:"Chapter 3 - Dimensions of Data Feasibility." National Academies of Sciences, Engineering, and Medicine. 2011. Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion. Washington, DC: The National Academies Press. doi: 10.17226/14509.
×
Page 18
Page 19
Suggested Citation:"Chapter 3 - Dimensions of Data Feasibility." National Academies of Sciences, Engineering, and Medicine. 2011. Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion. Washington, DC: The National Academies Press. doi: 10.17226/14509.
×
Page 19
Page 20
Suggested Citation:"Chapter 3 - Dimensions of Data Feasibility." National Academies of Sciences, Engineering, and Medicine. 2011. Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion. Washington, DC: The National Academies Press. doi: 10.17226/14509.
×
Page 20
Page 21
Suggested Citation:"Chapter 3 - Dimensions of Data Feasibility." National Academies of Sciences, Engineering, and Medicine. 2011. Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion. Washington, DC: The National Academies Press. doi: 10.17226/14509.
×
Page 21
Page 22
Suggested Citation:"Chapter 3 - Dimensions of Data Feasibility." National Academies of Sciences, Engineering, and Medicine. 2011. Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion. Washington, DC: The National Academies Press. doi: 10.17226/14509.
×
Page 22
Page 23
Suggested Citation:"Chapter 3 - Dimensions of Data Feasibility." National Academies of Sciences, Engineering, and Medicine. 2011. Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion. Washington, DC: The National Academies Press. doi: 10.17226/14509.
×
Page 23
Page 24
Suggested Citation:"Chapter 3 - Dimensions of Data Feasibility." National Academies of Sciences, Engineering, and Medicine. 2011. Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion. Washington, DC: The National Academies Press. doi: 10.17226/14509.
×
Page 24
Page 25
Suggested Citation:"Chapter 3 - Dimensions of Data Feasibility." National Academies of Sciences, Engineering, and Medicine. 2011. Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion. Washington, DC: The National Academies Press. doi: 10.17226/14509.
×
Page 25

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

C H A P T E R 3 Dimensions of Data FeasibilityTo determine the feasibility of using data from projects based on in-vehicle video data, the accuracy of two components of the data is critical. The components are (1) the video data and (2) parametric data, such as GPS data, radar-detected data, and complementary data. It is preferable to have continu- ously recorded and high-frequency data that are capable of showing the details of drivers’ maneuvers, environment, and vehicle status. The existence of complementary data and the availability of, for example, flow data, crash data, and weather data need to be investigated. An accurate time and location stamp that enables a proper link between vehicle data and com- plementary data is necessary to the vehicle and complementary data sources. Figure 3.1 shows the conceptual relationship between the data sets to be analyzed and the potential travel time reliability improvement measures. Some of the studies listed in Table 2.1 are readily suitable to serve the research purpose of this study. The quality of vehicle data and complementary data of each data set are dis- cussed individually in this section.Quality of Vehicle Data The following data sets were eliminated from the candidate list after examining the data size, quality, and availability. Project 1: Sleeper Berth The Sleeper Berth project instrumented two tractor trailers with cameras and DASs. Data from only 41 drivers were used in this study. Because of data acquisition limitations at the time the data were gathered, this data set is event-triggered. Consequently, it is not considered in the current study. Project 3: Quality of Behavioral and Environmental Indicators Used to Infer the Intention to Change Lanes This data set contains only 22 participants, and the data were collected in France. Because of the small sample size, potential differences in driver behavior relative to that in North Amer-15ica, and the challenge of using data from another country, this data set is considered unsuitable. Project 4: Lane Change FOT Because this is an older data set, only the most urgent lane change events have been converted to digital video. Significant effort and time would be required to digitize the data. Conse- quently, this data set is deemed unsuitable for the current study. Project 9: Naturalistic Driving Performance During Secondary Tasks This study used a small subset of data collected during the RDCWS FOT conducted by UMTRI. Because the RDCWS FOT is included in this research effort and is discussed in detail later, this subset of the data set is not considered further. Project 10: Effect of In-Vehicle Video and Performance Feedback on Teen Driving Behavior Instead of continuously recording data, the video cameras in this study were trigger-activated. Therefore, this data set is eliminated from further consideration. Project 12: CICAS-V Infrastructure The CICAS-V infrastructure study collected almost 1.5 TB of video and radar data for approaching vehicles at several instrumented signalized intersections. Driver image data were not collected in this study. The videos were installed at the intersections to capture vehicle movements and thus are not in-vehicle video data. Consequently, this data set is not considered further. Project 13: Pilot Study to Test Multiple Medication Usage and Driving Functioning According to the consent forms for the study, NHTSA is not allowed to share the video data with other parties.

16Figure 3.1. Relationship between project data sets and travel time reliability improvement.Consequently, the team would not have access to the data, and thus it is not considered further. Project 14: Older Driver FOT Data collection for this study is under way. It is anticipated that the data should be available in early 2010. Although the resulting data set has great potential to be used for future studies, it cannot be used in this study, given that the data are still being gathered. Project 15: CICAS-V Pilot FOT The CICAS-V pilot study has a relatively small data set. Only 87 drivers were recruited for this study, which involved driv- ing an experimental vehicle on a predetermined short route(approximately 40 mi). The driving data were not collected using a completely naturalistic method, and the length and roadways of the route are limited; therefore, this data set is excluded from the analysis. Project 16: Volvo Driving Behavior FOT This study commenced in May 2008 and is expected to last for 3 years. It will still be ongoing by the time this research effort ends. Therefore, this study is excluded. Data Sets After Initial Screening After the initial filtering, the resulting valid data sets are listed in Table 3.1. As can be seen from the table, data for Project 2 were collected in southeastern Michigan. For Project 5, data

17Table 3.1. Locations of Candidate Studies Candidate Data Set Dimensions MI DC DE MD NJ NY PA VA WV Project 2. ACAS FOT ✔ Project 5. RDCWS FOT ✔ Project 6. 100-Car Study ✔ ✔ ✔ Project 7. DDWS FOT ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ Project 8. NTDS ✔ ✔ ✔ ✔ ✔ ✔ ✔ ✔ Project 11. NTNDS ✔were collected in southeastern Michigan, including Detroit and surrounding suburbs and rural areas, resulting in 2,500 h of video data. For Project 6, data were collected in the Washing- ton, D.C., northern Virginia, and Maryland areas, resulting in 43,000 h of video data. For Project 7, drivers were recruited in Roanoke, South Boston, Stuarts Draft, and Cloverdale, Virginia, as well as in Charlotte, North Carolina. Data were collected for long-haul (cross-country) trips, as well as for overnight express (out-and-back) operations throughout the Mid-Atlantic area, resulting in 46,000 h of video data. For Project 8, drivers were recruited at trucking centers in Charlotte, Kernersville, and Henderson, North Carolina, and in Roanoke, Gordonsville, and Mount Crawford, Virginia. Data were collected for trucking runs throughout the Mid- Atlantic region, resulting in 14,500 h of video data. For Proj- ect 11, data were collected in the New River and Roanoke Valleys in Virginia, resulting in 16,644 h of driving data, including 10,754 h of teen driving data. Quality of External Data Using in-vehicle video data to help assess the role of driver behavior in nonrecurring congestion requires analyzing not only the vehicle and driver data but also the complementary data. For instance, it has been documented that weather affects driving behavior and performance and leads to longer follow- ing distances, thereby decreasing throughput at intersections and resulting in longer travel time. Another factor that has been shown to affect crash risk is the occurrence of a prior incident. Driver behavior in the vicinity of traffic control devices also contributes to nonrecurring congestion. Finally, previous studies in the Los Angeles conurbation have shown that more vehicle-hours of delay result from extraordinary and acciden- tally occurring traffic disturbances (nonrecurring) than from regularly occurring network overloading during typical daily peak hours (recurring). Although some of the complementary data can be obtained from data reduction (e.g., the traffic condition is a variable recorded by data reductionists while they were viewing videodata to describe the surrounding traffic), the availability of weather, traffic condition, crash, and work zone data in related states is investigated to provide a consistent data set across the studies and to avoid potential subjective bias brought by data reductions. Weather data can be reliably obtained by acquiring data from a nearby weather station. Figure 3.2 shows the locations of weather stations in the related states. As can be seen from the map, weather stations are densely located and thus there is a high possibility of linking a vehicle location to a nearby weather station through GPS data. Out of 592 weather stations, 253 are either Automated Surface Observing System (ASOS) stations or Automated Weather Observing System (AWOS) stations. Weather stations using ASOS are located at airports. ASOS is supported by the Federal Aviation Administration (FAA), the National Weather Service (NWS), and the Department of Defense (DOD). The system provides weather observations that include temperature, dew point, wind, altimeter setting, visibil- ity, sky condition, and precipitation. Five hundred sixty-nine FAA-sponsored and 313 NWS-sponsored ASOS stations are installed at airports throughout the United States. The weather reports by ASOS that could be used in this study are of METAR type (Aviation Routine Weather Reports) and contain precipi- tation type, precipitation intensities (in./h), and visibility read- ings (m). ASOS visibility measurements are performed at 30-s increments using a forward scatter sensor to compute 1-min average extinction coefficients (sum of the absorption and scat- tering coefficients). For this purpose, a photocell that identifies the time of day (day or night) is used to select the appropri- ate equation for use in the procedure. ASOS computes a 1-min average visibility level that is used to compute a 10-min moving average (MA) visibility level. This value is then rounded down to the nearest reportable visibility level. The system uses precipitation identification sensors to determine the type of precipitation (rain, snow, or freezing rain). Precipitation intensity is recorded as liquid-equivalent precipitation accumu- lation measurements using a Heated Tipping Bucket (HTB) gauge. The HTB has a resolution of 0.01 in. and an accuracy of ±0.02 in., or 4% of the hourly total, whichever is greater.

18Figure 3.2. Weather station locations.AWOS is one of the oldest automated weather stations and predates ASOS. It is a modular system utilizing a central processor that receives input from multiple sensors. Oper- ated and controlled by the FAA, state and local governments, and some private agencies, the AWOS reports weather infor- mation at 20-min intervals but does not report special obser- vations for rapidly changing weather conditions. Depending on the different varieties, AWOS observes different indices. The most common type, AWOS-III, observes temperature and the dew point in degrees Celsius, wind speed and direc- tion in knots, visibility, cloud coverage and ceiling up to 12,000 ft, and altimeter setting. Additional sensors, such as for freezing rain and thunderstorms, have recently become available. Traffic count data are available in all the states that were studied. The state of Virginia has extensive locations of traf- fic count stations. There are more than 72,000 stations, 470 of which collect continuous data. A subset from the Virginia Count Station list—the traffic count locations in the city of Richmond—is shown in Figure 3.3. The traffic count stations in West Virginia are shown in Figure 3.4. There are approx- imately 60 permanent count stations in West Virginia, and one-third of the state is counted each year in a CoverageCount Program. In Pennsylvania, there are 116 continuous stations out of 30,661 count stations. Figure 3.5 shows a sub- set of the traffic count stations in Pittsburgh. Figure 3.6 shows traffic count stations in Delaware. A sample of the longitudinal and latitudinal information of one traffic count station on link ID 507002 is shown in Table 3.2. Table 3.3 demonstrates a sample of the raw traf- fic counts collected by that station at 15-min intervals by vehicle class in the state of Virginia. Traffic conditions (e.g., traffic density and level of service) can be inferred from the counts. The longitudinal and latitudinal fields can then be used to link in-vehicle GPS data and traffic count data. Some states have location information for the stations listed as mileposts and street names that can be digitized when necessary. Crash data are readily available for every state, although some have stringent data privilege requirements. Some states, such as New Jersey and Michigan, have online crash databases from which the information can be downloaded. The District of Columbia DOT coded their crashes with work zone infor- mation if there was a work zone in the surrounding area when the crash happened. Table 3.4 provides a crash sample from Washington, D.C. Because of space limitations, only

Figure 3.3. Traffic count stations in Richmond, Va.Figure 3.4. Traffic count stations in West Virginia.

20Figure 3.5. Traffic count stations in Pittsburgh, Pa.some of the variables in the original database are listed here. Other information, such as road condition, light condition, weather, and the sobriety of involved drivers, is listed in the original database. Crash and traffic volume data are typically saved in a data- base, but work zone data usually are not. This is especially true for completed road work. Most states have ongoing projects recorded in the database, but completed projects are not stored. A few states have incomplete data or data that are not in suffi- cient condition for use. For example, as of August 2008 in the state of Virginia, the 511 system has used a program called VA Traffic to log information. Before August 2008, there was lim- ited free-text-based information recording the scheduled start and end of road work. For the state of Pennsylvania, the work zone data are only available for projects that occur on state high-ways. Table 3.5 provides a sample of the work zone data from West Virginia. Table 3.6 summarizes the availability of complementary data in related states. Online sources from which data can be downloaded are listed in the footnotes.Evaluation of Candidate Data Sets To help determine the feasibility of candidate databases, a multidimensional criterion is established for the data sources. These dimensions include comprehensiveness, video data quality, vehicle data, linkages, and data format and structure. Table 3.7 provides a detailed explanation and definition for each feasibility dimension.

21Figure 3.6. Traffic count stations in Delaware. Color version of this figure: www.trb.org/Main/Blurbs/165281.aspx.Table 3.2. Location of Traffic Station Link ID Counter Sensor Physical Location Latitude Longitude 507002 1 1 0.205 mi from Country Club 37.21934 −80.4056Each candidate database is evaluated on each dimension to demonstrate the suitability for further analysis. At the same time, the legal restriction for each data set is examined. Cer- tain data sets have IRB restrictions, meaning that the data col- lected in that study are restricted for external usage and need to be eliminated from the candidate pool. Some studies col- lected video data at a lower frequency and, therefore, are not suitable for this study.To quantitatively evaluate each qualified candidate data- base, a composite feasibility score is computed to reflect the database’s strengths and weaknesses, as displayed in Table 3.8. Each dimension receives a seven-point scale score, ranging from 1, representing a low score (e.g., small sample and data available only in raw form), to 7, representing a high score (e.g., large, representative sample and data in reduced form). In computing the score, each feasibility category is assigned a weight so that the sum of weights across all cat- egories totals 100 (e.g., the weight for the feasibility cate- gory comprehensiveness is 15). Within each feasibility category the various measures that constitute a category are assigned a score so that the sum of scores within a category is 10. For example, the comprehensiveness category includes four measures: (1) driver population, (2) types of road- ways, (3) types of trips, and (4) types of vehicles. These mea- sures are assigned a weight of 4, 4, 1, and 1, respectively. The weights are used to compute a weighted average score for each feasibility category. The feasibility category scores are then used to compute a weighted average overall score between 1 and 7. The quality of video data is vital to this project. Some dimensions, such as whether the driver’s face and hand movements can be clearly seen or whether the sur- rounding vehicles can be accurately located by the radar sen- sors, receive greater weights to emphasize the importance of those data to this study. To further illustrate the scoring methodology, the pro- cedure is demonstrated using the 100-Car data set. The score for the comprehensiveness category is computed as the weighted average of the four measures that constitute this category as This computation is repeated for each of the remaining feasibility categories (video data quality, vehicle data, link- ages, and data format and structure). The overall score is then computed as the weighted average of the various cat- egory scores as 6.50 100 × + × + × + × + × = 15 6 20 40 6 11 20 2 67 20 7 00 5 5 . . . . . ( )6 2 6 10 × + × + × + × = 4 7 4 6 1 7 1 6 50 1. ( )

22Table 3.3. Sample of Traffic Count Data in Virginia Link ID Direction Lane Start Date and Time Interval Class Quality Class 15 507002 1 1 3/15/2007 7:00 15 1 4 507002 1 1 3/15/2007 7:15 15 1 6 507002 1 1 3/15/2007 7:30 15 1 20 507002 1 1 3/15/2007 7:45 15 1 26 507002 1 1 3/15/2007 8:00 15 1 20 507002 1 1 3/15/2007 8:15 15 1 26 507002 1 1 3/15/2007 8:30 15 1 32 507002 1 1 3/15/2007 8:45 15 1 20 507002 1 1 3/15/2007 9:00 15 1 22 507002 1 1 3/15/2007 9:15 15 1 8 507002 1 1 3/15/2007 9:30 15 1 10 507002 1 1 3/15/2007 9:45 15 1 10 507002 1 1 3/15/2007 10:00 15 1 6 507002 1 1 3/15/2007 10:15 15 1 4 507002 1 1 3/15/2007 10:30 15 1 7 507002 1 1 3/15/2007 10:45 15 1 11 507002 1 1 3/15/2007 11:00 15 1 8 507002 1 1 3/15/2007 11:15 15 1 6 507002 1 1 3/15/2007 11:30 15 1 10 507002 1 1 3/15/2007 11:45 15 1 18 507002 1 1 3/15/2007 12:00 15 1 4 507002 1 1 3/15/2007 12:15 15 1 16 507002 1 1 3/15/2007 12:30 15 1 18 507002 1 1 3/15/2007 12:45 15 1 12 507002 1 1 3/15/2007 13:00 15 1 18 507002 1 1 3/15/2007 13:15 15 1 9Table 3.4. Crash Sample Data from Washington, D.C. No. of No. of No. of No. of Passengers Passengers Date Time Report Type Street Block Vehicles Injuries in Car 1 in Car 2 01/22/07 19:20 1/22/07 Injury Good Hope Rd. 2300 2 1 1 2 01/22/07 19:20 1/22/07 Prop. Damage Benning Rd. 3330 1 0 1 0 01/23/07 12:20 1/23/07 DC Property Benning Rd. 4500 2 0 1 1 08/12/07 11:25 8/12/07 Injury Southern Ave. 4400 2 1 1 0 08/12/07 14:00 8/12/07 Hit and Run 57th St. 100 2 0 1 0

23Table 3.5. Sample Work Zone Data from West Virginia County Route Project Description Miles Start Date Completion Date Braxton WV 4 Replace Drainage, Gassaway–Sutton Road 0.68 9/25/2008 9/28/2008 Braxton I-79 Reset Bearings, Sutton/Gassaway Bridge 0.22 5/28/2008 7/28/2008 Braxton I-79 Resurfacing, County 19/26–Flatwoods 2.76 6/28/2008 10/31/2008 Braxton I-79 Resurfacing, Flatwoods–Burnsville Road 3.74 5/28/2008 10/31/2008 Braxton I-79 Resurfacing, WV 5–Burnsville 3.95 6/28/2008 10/31/2008 Brooke WV 2 Beech Bottom–Wellsburg Road 4.32 7/28/2008 10/28/2008 Brooke WV 2 Follansbee–Coketown Road 2.09 7/1/2008 10/28/2008Table 3.6. Environmental Data of Candidate Studies Traffic Work Crash State Count Zone Log Log Virginia Yes Yesa Yes Delaware Yes No Yes Maryland Yes No Yes Washington, D.C. Yesb Yesc Yes New Yorkd Yese Yes Yes New Jersey Yesf No Yesg Pennsylvaniah Yes Yes Yes West Virginia Yesi Yesj Yesk Michigan Yes No Yes aData before August 2008 incomplete. bData available at http://ddot.dc.gov/DC/DDOT/About+DDOT/Maps/Traffic+ Volume+Maps. According to DDOT, the data were collected using portable counters every 3 years and are converted to Annual Average Daily Traffic (AADT) as shown on the map. cData available online at http://app.ddot.dc.gov/information_dsf/construction/ index.asp?pro=COM&wardno=1. dOnline real-time data available at www.511ny.org/traffic.aspx. eOnline data available at http://gis.nysdot.gov/tdv/. fOnline data available at www.state.nj.us/transportation/refdata/roadway/traffic_ counts/. gOnline data available at www.state.nj.us/transportation/refdata/accident/ rawdata01-03.shtm. hOnline real-time data available at www.dot7.state.pa.us/TravelerInformation/. iAADT data online at www.wvdot.com/3_roadways/rp/TA%20Traffic%20files/ SecBcounts.htm. jReal-time work zone information online www.transportation.wv.gov/highways/ traffic/Pages/roadconditions.aspx. Historical work zone data available for the past 2–3 years. kwww.transportation.wv.gov/highways/traffic/Pages/default.aspx.The results show that the NTDS and the NTNDS score the highest, with high scores for video, vehicle, and format struc- ture measures. Although these data sets are limited in terms of population or vehicle coverage, the weights assigned to these categories were lower; thus the final score is still consid- ered high relative to the other studies. The next two studies are the 100-Car Study and the DDWS FOT. Noteworthy is the fact that the truck and teen driver studies scored low in the comprehensiveness category because they are restricted either to specific vehicle types or specific driver populations. Given that the focus of this project is to investigate the feasi- bility of using video data to characterize driver behavior before nonrecurring congestion, this feasibility category is not assigned a high weight. All the data sources are accessible to the research team, although some limitations may apply. For example, the teen study conducted by VTTI, which studied minors, would require special procedures before any data mining could be conducted. Specifically, data for this study are strictly limited to VTTI researchers, and data reduction can be conducted only in a separate laboratory in which data reductionists can- not be seen by other personnel. Special instructions should be given to reductionists regarding what to do if they find sensi- tive data or if they meet participants socially and other such conduct instructions. The resulting qualified data sets are listed in Table 3.8. As can be seen from the table, Projects 6, 7, 8, and 11 score relatively higher than the other projects overall. These projects collected data with fewer flaws in video data. Postprocessing of that data will require fewer monetary and human resources.

24Table 3.7. Definitions of Dimensions Feasibility Dimensions Definitions Institutional Review Board (IRB) Comprehensiveness Driver population Types of roadways Types of trips Types of vehicles Video Data Quality Driver’s hand and foot movements captured Driver’s face captured Front view Side view Rear view Vehicle Data Lane location for each target Projected collision time Speed Headway Accelerometer Braking GPS Lane-changing behavior Lateral placement Linkages Ability to link to environmental data Ability to link to operational data Ability to link to special-event data Ability to link to incident data Ability to link to traffic control devices Ability to link to work zone data Data Format and Structure Sampling rate suitability Event or continuous Reduced or raw Do the consent forms used in the study allow for the data to be released to third parties? Does the data set contain a sample of heterogeneous drivers (e.g., teens, older adults, novices)? Are multiple driving locations (freeways, urban settings, rural roads) represented in the data set? Does the data set contain different trips that occurred at different times during the day (e.g., peak and nonpeak)? Does the data set contain multiple vehicle types? Can the driver’s hands and feet be clearly seen in the video? Does the video resolution allow for eyeglance reduction? Is it possible to see driver–passenger interactions and other sources of distraction? Do the camera views allow verification of interaction between the vehicle and other vehicles in sight? Do camera views outside the vehicle allow the researcher to see what the driver is responding to by the side of the vehicle? Will the following vehicle be seen in the video? Are radar data available for calculating lane locations for other targets? Is TTC available in the data set or can it be calculated by data reductionists? Is vehicle speed available? Is headway available (either distance or time headway)? Is acceleration measured? Is braking behavior recorded in the data? Are GPS data available to identify the vehicle’s location? Is lane-changing behavior coded in the data or in the reduced data? Is the lateral placement of the vehicle measured? Is it possible to link the data to environmental data (such as weather) using the time and location information? Will it be possible to link the vehicle data with the surrounding operational data, such as traffic volume or congestion? Is the time stamp valid to link the data to a surrounding special event? Are the crash data available to be linked to the data sets? Can any traffic control devices (e.g., traffic light, stop sign, yield sign) be linked to the data set? Can work zone data be linked to the data set? Is the sampling rate of the data collection sufficient to understand driver behavior and traffic conditions outside the vehicle? Does the data set contain continuous driving behavior, or just segments that are event-triggered? Is the data set already in a format that would allow for efficient analysis (reduced), or are the data only available in a raw format?

25Table 3.8. Scale Scores of Candidate Studies Project 5: Road Project 7: Project 2: Departure Drowsy Project 8: Project 11: ACAS Crash Driver Naturalistic Naturalistic Field Warning Project 6: Warning Truck Teen Operational System 100-Car System Driving Driving Feasibility Dimensions (Score) Weight Test (FOT) FOT Study FOT Study Study Legal Restrictions (0/1) 1 1 1 1 1 1 Comprehensiveness 15 6.40 6.40 6.50 3.40 3.80 5.30 Driver population 4 7 7 6 4 5 3 Types of roadways 4 7 7 7 4 4 7 Types of trips 1 7 7 6 1 1 6 Types of vehicles 1 1 1 7 1 1 7 Video Data Quality 40 3.40 3.40 6.20 7.00 7.00 6.20 Driver’s hand and foot movements captured 2 1 1 7 7 7 7 Driver’s face captured 2 7 7 7 7 7 7 Front view 2 7 7 7 7 7 7 Side view 2 1 1 3 7 7 3 Rear view 2 1 1 7 7 7 7 Vehicle Data 20 7.00 5.33 6.11 6.00 7.00 6.33 Lane location for each target 1 7 4 5 7 7 5 Projected collision time 1 7 1 7 7 7 7 Speed 1 7 7 7 7 7 7 Headway 1 7 1 7 7 7 7 Accelerometer 1 7 7 7 7 7 7 Braking 1 7 7 7 7 7 7 GPS 1 7 7 5 4 7 7 Lane-changing behavior 1 7 7 5 7 7 5 Lateral placement 1 7 7 5 1 7 5 Linkages 20 5.5 5.5 2.67 3.17 5.50 5.83 Ability to link to environmental data 2 7 7 2 3 7 7 Ability to link to operational data 2 7 7 2 3 7 7 Ability to link to special-event data 2 4 4 2 4 4 4 Ability to link to incident data 2 7 7 2 3 7 7 Ability to link to traffic control devices 2 6 6 6 3 6 6 Ability to link to work zone data 2 2 2 2 3 2 4 Data Format and Structure 5 5.20 5.20 7.00 7.00 7.00 7.00 Sampling rate suitability 1 7 7 7 7 7 7 Event or continuous 6 7 7 7 7 7 7 Raw or reduced 3 1 1 7 7 7 7 Overall Score (1–7) 5.1 4.7 5.6 5.5 6.2 6.1

Next: Chapter 4 - Develop a Methodology for Analyzing Data »
Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s second Strategic Highway Research Program (SHRP 2) Report S2-L10-RR-1: Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion presents findings on the feasibility of using existing in-vehicle data sets, collected in naturalistic driving settings, to make inferences about the relationship between observed driver behavior and nonrecurring congestion.

The report, a product of the SHRP 2 Reliability focus area, includes guidance on the protocols and procedures for conducting video data reduction analysis.

In addition, the report includes technical guidance on the features, technologies, and complementary data sets that researchers can consider when designing future instrumented in-vehicle data collection studies.

The report also highlights a new modeling approach for travel time reliability performance measurement across a variety of traffic congestion conditions.

An e-book version of this report is available for purchase at Google, Amazon, and iTunes.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!